Research map / 3D vision / 3D generation
“What Happens If...” Learning to Predict the Effect of Forces in Images
Roozbeh Mottaghi, Mohammad Rastegari, Abhinav Gupta, Ali Farhadi
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add depth channel to image tower | ours w/ ResNet · ForScene sequence prediction (test) | sequences correct (%): 16.9 -> 19.8 | Improved |
| add depth channel to image tower | ours w/ AlexNet · ForScene sequence prediction (test) | sequences correct (%): 16.5 -> 17.5 | Improved |
| replace AlexNet image tower with ResNet-18 | ours · ForScene sequence prediction (test) | sequences correct (%): 16.5 -> 16.9 | Improved |
| add RNN sequence model instead of fully connected regression (ablation) | ours w/ AlexNet · ForScene sequence prediction (test) | sequences correct (%): 10.4 -> 16.5 | Improved |
| add RNN sequence model instead of nearest neighbor retrieval (ablation) | ours w/ AlexNet · ForScene sequence prediction (test) | sequences correct (%): 8.5 -> 16.5 | Improved |
| relax criterion to edit distance <= 1 | ours w/ AlexNet · ForScene sequence prediction (test) | sequences correct (%): 16.5% -> 59.8% | Improved |
| train with all categories instead of holding one out (ablation) | ours w/ AlexNet · ForScene unseen categories (leave-one-category-out) | sequences correct (%): 16.12 -> 16.53 | No clear effect |
Closest papers by meaning
- Visual Interaction Networks: Learning a Physics Simulator from Video (NeurIPS 2017 · similarity 0.87)
- Physion++: Evaluating Physical Scene Understanding that Requires Online Inference of Different Physical Properties (NeurIPS 2023 · similarity 0.86)
- Newtonian Scene Understanding: Unfolding the Dynamics of Objects in Static Images (CVPR 2016 · similarity 0.86)
- Use the Force, Luke! Learning to Predict Physical Forces by Simulating Effects (CVPR 2020 · similarity 0.85)
- Bounce and Learn: Modeling Scene Dynamics with Real-World Bounces (ICLR 2019 · similarity 0.85)
- Learning Long-term Visual Dynamics with Region Proposal Interaction Networks (ICLR 2021 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
